Calibração e validação experimental do modelo numérico de um vagão de carga em condições operacionais

Abstract

With the advancement of rail freight in the world and the consequent increase in speed and load transported by compositions, problems associated with the dynamics of rail vehicles have become increasingly frequent, requiring increasingly accurate methods to evaluate their effects. Therefore, this work presents an efficient methodology for the calibration and validation of a numerical model of a freight wagon using an iterative methodology, through a genetic algorithm and based on modal parameters experimentally identified in a dynamic test under real operating conditions. The dynamic tests involved the use of an onboard monitoring system, composed of a set of accelerometers and linear variable displacement transducers (LVDT's). The number and location of the sensors was severely conditioned by the space occupied by the transported load, as well as by restrictions associated with the loading and unloading of the wagon, thus, no interference with the vehicle operation was created. The data derived from the dynamic test were used, first, to identify the modal parameters of the carbody, that is, the natural frequencies, forms of vibration and damping coefficients, and second, to extract the time histories of accelerations and displacements in conditions of operation. The identification of dynamic properties was made through the application of operational modal analysis techniques to the data collected during vehicle circulation. The decision to adopt a minimalist experimental setup, which would not impact vehicle operation, resulted in extra difficulties for modal identification. However, even under unfavorable conditions, the first three rigid body modes of the carbody were identified, which served as the basis for model calibration. A simplified numerical finite element model of the freight wagon was developed and calibrated using an iterative methodology through a genetic algorithm and based on the identified modal parameters, which was implemented with the interaction between MATLAB® and ANSYS® software. As with modal identification, the limited experimental information imposed some extra difficulties on the calibration process, especially regarding the pairing between numerical and experimental modes. Even with these limitations, the methodology applied proved to be effective and robust in accurately estimating three numerical parameters, in addition to providing significant improvements in relation to the numerical model before calibration These improvements were confirmed by the reduction from 8.71 % to 0.03 % of the average percentage difference between numerical and experimental natural frequencies. Finally, the dynamic response of the model was validated by means of a direct comparison between numerically simulated results, based on a dynamic vehicle track interaction analysis, and experimentally collected time history responses. Comparisons were made with the model before and after the calibration, through which excellent agreement was obtained between the experimental and numerical time series after the calibration of the model. Thus, it was concluded that the calibration process was able to significantly improve the numerical model.